Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Theoretical and Empirical Minimum Detectable Displacements for
Deformation Networks
Bahattin ERDOGAN, Serif HEKİMOGLU, Utkan Mustafa DURDAG*, Turkey
Key words: GPS Network, Horizontal Control Network, Displacement Ellipse, Theoretical
Minimum Detectable Displacement, Empirical Minimum Detectable Displacement
SUMMARY
In deformation analysis, statistical methods have been used to detect displaced point(s) and to
identify the characteristic feature of object with a high of probability. These methods can
sometimes have wrong results; that’s why the reliability of the method should be known.
However; to measure the reliabilities of the methods, it is required to know actual displaced
point(s) before the analysis. Since it is impossible in practice, displacements of points are
simulated. Displacement ellipse of point have been used to determine the magnitude of
minimum detectable displacements of the point. However, the magnitudes and directions of
semi-minor and semi-major axes are different, corresponding displacement circle can be used to
calculate minimum detectable displacement. Displacement circle is an empirical method. Also,
minimum detectable displacement based on the non-centrality parameter is also important for the
sensitivity of a deformation network which is characterised its efficiency to detect displacements
in the area covered by the network. This method is called theoretical method. In this study, we
have investigated for answering the question “which method is more realistic?”. Both empirical
and theoretical minimum detectable displacements have been calculated for horizontal control
network and GPS network. At the same time, an empirical method based on global test for
minimum detectable displacement has been estimated. The theoretical minimum detectable
displacements changes for different power of the test value (e.g. 70%, 80%). It has been shown
that the theoretical minimum detectable displacements converges to empirical minimum
detectable displacements when the power of test is %70.
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
Theoretical and Empirical Minimum Detectable Displacements for Deformation
Networks
Bahattin ERDOGAN, Serif HEKİMOGLU, Utkan Mustafa DURDAG*, Turkey
1. INTRODUCTION
Being the cause of massive losses of life and property, deformation analysis is of vital importance
in determining the movements originated from tectonics, landslides or man-made buildings (dams,
bridges etc.). Conventional vertical, horizontal or three dimensional networks as well as GPS
networks are established to control behaviour of these movements periodically. The main goal is to
answer the question that whether any displacement occur or not. Various studies have been
performed regarding deformation analysis (Pelzer, 1971; Caspary et al, 1983; Koch, 1985;
Niemeier, 1985; Schaffrin, 1986; Caspary, 1988; Caspary et al., 1990; Schaffrin & Bock, 1994;
Betti et al., 1999; Snow & Schaffrin 2007; Cai et al., 2008).
Accuracy, reliability and sensitivity should be taken into consider in the geodetic deformation
network design phase (Cooper, 1987). Accuracy refers to quality of the network. Reliability is the
meaning of effect of the undetectable outliers on deformation analysis (Aydin & Demirel, 2005).
Detectability of expected displacements and deformations at the designated network corresponds to
sensitivity (Even-Tzur, 2006). To measure the efficacy of the Conventional Deformation Analysis
(CDA), prior knowledge about which point has been displaced is needed. In practice, it is
impossible to obtain this information; hence the points’ displacements have been simulated for
various studies (Hekimoglu et al., 2010; Duchnowski & Wisnievski, 2014; Nowel & Kaminski,
2014; Velsink, 2015; Nowel, 2015).
There are three main factors affect the sensitivity of deformation network; network design, session
duration (for GPS measurements) and accuracy of measuring instrument. Since the accuracy of
estimated coordinates and their (co)variances depends on the all these factors, the network should
be capable of detecting expected deformation magnitude. To determine the detectable displacement
magnitude, the ellipses of the points are used as criterion. Displacement area is identified with the
displacement ellipses. Standardized expected displacement ellipses of the points in the network can
be used. Displacement ellipses are the global confidence region. If the magnitude of the estimated
displacement vector lies entirely within the properly scaled expected point displacement ellipse,
then there is no significant movement of the point at the chosen significance level α (Cooper, 1987).
For the simulation, one main problem is how to generate the displacement magnitude in the
deformation networks. In this study, our aim is to specify Theoretical and Empirical Minimum
(Minimal) Detectable Displacements (TMDD, EMDD). These magnitudes have been used to
compare with each other for Horizontal Control Network (HCN) and also GPS network. For this
purpose; 3 points (A, B and C) of HCN and 3 points of GPS network (OBC1, OBC2 and OBC3)
were chosen to calculate one TMDD and two EMDD’ s magnitudes.
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
Overview of this paper: After a review of the conventional deformation approach in Sect. 2, the
paper gives in Sect. 3 a description of the EMDD. Sect. 4 presents TMDD and simulated networks
have shown in Sect. 5. Conclusions are given about the results of the experiments in Sect. 6.
2. CONVENTIONAL DEFORMATION ANALYSIS
The typical CDA compares coordinate differences in a geodetic network between two different
observation epochs. If H0 hypothesis rejected as expected movements by statistical tests, these
differences are interpreted as being the “displacements.” Conventionally, the object and its
surrounding are characterised by discrete points in deformation monitoring network and only the
geometrical changes of the object are modelled. The monitoring network is adjusted as a free
network for each epoch and coordinate differences between two different observation epochs are
compared. The first step of the CDA is global congruency test. The model which suggests that the
coordinates of corresponding points between two epochs do not change differently from
expectations is formed as the null hypothesis (Welsch & Heunecke 2001)
𝐻0: [−𝐈 𝐈] [𝐸(�̂�𝟏)
𝐸(�̂�𝟐)] = 𝟎 (1a)
𝐻1: [−𝐈 𝐈] [𝐸(�̂�𝟏)
𝐸(�̂�𝟐)] ≠ 𝟎 (1b)
where E() stands for expectation, �̂�𝟏 and �̂�𝟐 : The vectors of estimated coordinates of the first and
the second epochs, �̂�𝟏 and �̂�𝟐 are estimated by free network adjustment as following equations:
�̂�𝟏 = 𝐐�̂�𝟏�̂�𝟏𝐀𝟏𝐓𝐏𝟏𝐥𝟏, 𝐐�̂�𝟏�̂�𝟏 = 𝐍𝟏𝟏
+ = (𝐀𝟏𝐓𝐏𝟏𝐀𝟏)+ (2)
�̂�𝟐 = 𝐐�̂�𝟐�̂�𝟐𝐀𝟐𝐓𝐏𝟐𝐥𝟐, 𝐐�̂�𝟐�̂�𝟐 = 𝐍𝟐𝟐
+ = (𝐀𝟐𝐓𝐏𝟐𝐀𝟐)+ (3)
Ω = 𝐯𝟏𝐓𝐏𝟏𝐯𝟏 + 𝐯𝟐
𝐓𝐏𝟐𝐯𝟐 (4)
𝑠02 =
Ω
𝑓, 𝑓 = 𝑓1 + 𝑓2 (5)
where 𝐀𝟏 and 𝐀𝟐 : Design matrices of the first and second epochs, respectively, if the configuration
of the network in different epochs is not changed 𝐀𝟏=𝐀𝟐, �̂�𝟏 and �̂�𝟐 : Vectors of estimated
coordinates of the first and the second epochs, 𝐏𝟏 and 𝐏𝟐 : Weights matrices at the first and the
second epochs, 𝐥𝟏 and 𝐥𝟐 : Observations vectors of the first and the second epochs, 𝐯𝟏 and 𝐯𝟐 :
Residual vectors of the first and the second epochs, 𝐟𝟏 and 𝐟𝟐 : Degrees of freedoms of the first and
the second epochs, respectively.
Then, the influence of the null hypothesis on the Least Square Estimation (LSE), in the absence of
correlations between epochs, results in (Pelzer 1971; Niemeier 1985; Koch 1985; Cooper 1987)
𝐝 = �̂�𝟐 − �̂�𝟏 (6)
𝐐𝐝𝐝 = 𝐐�̂�𝟏�̂�𝟏 + 𝐐�̂�𝟐�̂�𝟐 (7)
𝑅 = 𝐝𝐓𝐐𝐝𝐝+ 𝐝 (8)
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
where d : difference vector of estimated coordinates, 𝐐𝐝𝐝 : Cofactor matrix of d, 𝐐𝐝𝐝+ :
Pseudoinverse of the cofactor matrix.
𝑇 =𝑅
ℎ𝑠02 ~𝐹ℎ,𝑓,𝑎, under 𝐻0, assuming normally distributed observations errors,
h is the rank of the matrix 𝐐𝐝𝐝, 𝑎 : Error probability, 𝑠02 : Estimated variance component in the
absence of the null hypothesis.
If 𝑇 >𝐹ℎ,𝑓,𝑎, then the null hypothesis is rejected. Thus, the difference in the coordinates between two
epochs is interpreted as the result of an unexpected displacement (Niemeier, 1985; Welsch &
Heunecke 2001).
3. EMPIRICAL MINIMUM DETECTABLE DISPLACEMENT
The lengths and directions of the semi-axes of the expected displacement ellipses are different for
each point in the network. Thus these ellipses are not convenient for a simulation which generates
the magnitude of the displacement. To compare the successes of the CDA in different cases, a circle
is chosen instead of an ellipse. The circle whose area is equal to the area of the expected
displacement ellipse could be chosen as given in Fig. 1, so that the total area of the positive part is
equal to the total area of the negative part. The radius of the displacement circle may be estimated
as a mean value from the different radius for a given network (Hekimoglu et al., 2010). First EMDD
magnitude calculated from the formulas shown as in the following section.
3.1 Empirical Technique 1: Using Displacement Ellipse
Fig. 1. Expected displacement ellipse and corresponding displacement circle of Point P (a and b are
the semi-major and semi-minor axes of the expected displacement ellipse) (Hekimoglu et al., 2010).
The formulas for the elements of the rescaled expected displacement ellipse and the corresponding
displacement circle are given as follows:
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
int = 𝑞1𝑟 < |𝑧𝑖| < 𝑞2𝑟, 𝑞1 ≥ 1, 𝑞1 ≠ 𝑞2, 𝑞1 < 𝑞2 < 3 (9)
𝑎 = 𝜎0√𝜆1𝜒2,1−𝛼2 (10)
𝑏 = 𝜎0√𝜆2𝜒2,1−𝛼2 (11)
𝜆1 =𝑄𝑑𝑥𝑖𝑥𝑖+𝑄𝑑𝑦𝑖𝑦𝑖+𝑤
2 (12a)
𝜆2 =𝑄𝑑𝑥𝑖𝑥𝑖+𝑄𝑑𝑦𝑖𝑦𝑖−𝑤
2 (12b)
𝑤 = √(𝑄𝑑𝑥𝑖𝑥𝑖 − 𝑄𝑑𝑦𝑖𝑦𝑖)2 + 4𝑄𝑑𝑥𝑖𝑦𝑖2 (13)
𝑟 = √𝑎𝑚𝑒𝑎𝑛𝑏𝑚𝑒𝑎𝑛 (14)
where 𝑄𝑑𝑥𝑖𝑥𝑖 and 𝑄𝑑𝑦𝑖𝑦𝑖 = elements of the respective submatrix of the cofactor matrix 𝐐𝐝𝐝 from Eq.
7, which belongs to the ith
point, 𝜆1 and 𝜆2 = Semi-major and semi-minor axes of the standardized
expected displacement ellipse, 𝜎02 = A priori variance component, 𝜒2,𝑎
2 = 𝑎 - fractile of the 𝜒22-
distribution for 2 degrees of freedom, r = Radius of the corresponding displacement circle.
The value of 𝑎 is chosen here as 0.001 so the stochastic effect is reflected almost entirely in the
simulated displacement magnitude.
3.2 Empirical Technique 2: Step Approach Testing
Second EMDD magnitude was obtained by step approach testing. This technique estimates the
minimum magnitude of displacement for different directions depending on the global congruency
test that is given in Eqs. (6), (7) and (8). To reach the final step quicker, the first value has chosen
randomly which can both detectable and reflects the stochastic model of network. Then first value
was increased or decreased 0.1 mm to get the minimum displacement magnitude.
3.2.1 Simulation of Displacement Vector
To use in deformation analysis, the random errors were generated differently for each epoch. These
vectors were added to free-error measurements. Only random errors (e1) were added to 1st epoch
measurements. For 2nd
epoch; both random errors (e2) and displacements of the points (𝐳) were
added to free-error measurements. 1st and 2
nd epoch measurements given as follows (Hekimoglu et
al., 2010):
𝐥𝟏 = 𝐥̅ + 𝐞𝟏 (15)
𝐥𝟐 = 𝐥̅ + 𝐞𝟐 + 𝐀𝐳 (16)
Where e1 and e2 : normally distributed random error vectors, A : Coefficient matrix, 𝐳 : Horizontal
displacement vector and 𝐥 ̅: the observation vector without random errors.
𝐳 = [𝑧1𝑥 𝑧1𝑦 𝑧2𝑥 𝑧2𝑦 … 𝑧𝑢𝑥 𝑧𝑢𝑦] (17)
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
zx : The projection of displacement vector z to x axes of the displacement, zy : The projection of
displacement vector z to y axes of the displacement.
4. THEORETICAL MINIMUM DETECTABLE DISPLACEMENT
The minimal detectable value can be used as a different approach to obtain the test value. Let dk and
𝜎02 be the expected displacement vector and a priori variance of unit weight, respectively. To
determine these displacemets the alternative hypothesis must be accepted as shown in the Eq. 1b.
Then the test value is non-central F distribution and its eccentricity parameter is given as following
equation:
𝜆 =𝐝𝐤
𝐓𝐐𝐝𝐝+ 𝐝𝐤
𝜎02 (18)
The eccentricity parameter for specific value, power of test 𝛾 = 1 − 𝛽, error probability α and
degrees of freedom (h,𝑓 = ∞), is compared with the threshold value 𝜆0. If 𝜆 > 𝜆0 it is inferred that
the deformation network is sensitive. This comparison process called as sensitivity analysis
(Caspary et al., 1983; Niemeier, 1985; Cooper, 1987; Even-Tzur, 2006; Aydin et al., 2004).
To determine the minimal detectable displacement magnitude of the points of network vector g that
consists of movement direction is generated. For the two dimensional network the vector g is as
𝐠 = [cos 𝑡1 sin 𝑡1 cos 𝑡2 sin 𝑡2 … cos 𝑡𝑝 sin 𝑡𝑝]𝑇 (19)
where p : The number of points, in vector g the components of the points which are considered as
undisplaced is assumed as “0”. Thus the deformation vector is as
𝐝𝐤 = a𝐠 (20)
where a : The scale factor and it is computable. Then, by substituting Eq. (20) into Eq. (18) the
equation has formed as
𝜆 =a2
𝜎02 𝐠𝐓𝐐𝐝𝐝
+ 𝐠 (21)
Then, by considering the inequation of 𝜆 > 𝜆0 with Eq. (21) the Eq. (22) is obtained as follows:
a2
𝜎02 𝐠𝐓𝐐𝐝𝐝
+ 𝐠 > 𝜆0 (22)
From Eq. (22) the minimum value of “a” can be calculated as follows:
a𝑚𝑖𝑛 = 𝜎0√𝜆0
𝐠𝐓𝐐𝐝𝐝+ 𝐠
(23)
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
From the Eq. (20) the minimum detectable displacement magnitude can be obtained at any desired
point in any direction by using amin.
5. SIMULATED NETWORKS
5.1 Horizontal Control Network
Horizontal Control Network was simulated shown as Fig. 2. The number of observations and the
number of point are 72 and 9, respectively. The degrees of freedom is 48 (by taking 9 orientation
unknowns into account) with regard to free network adjustment. The lengths of the distance
measurements are varied approximately between 3 and 7 km. Random error vectors were generated
from a normal random error generator. The random errors 𝑒𝑗𝑘 come from the different normal
distributions 𝑁(𝜇 = 0, 𝜎𝑑2), where 𝜎𝑑 was chosen to be the same for the direction measurements at
each point, such as ±0.3 mgon and 𝜎𝑑 = ±3+2ppm mm for the distance measurements.
Fig. 2. Horizontal Control Network (Erdogan 2011, Hekimoglu et al., 2010).
The two epochs’ geometries are the same; so the weight matrices of the observations are equal to
each other for the two epochs as follows (assuming no correlation between epochs):
𝐏 = [𝐏𝐝 𝟎𝟎 𝐏𝐬
] (24)
Where 𝐏𝐝 = 𝑑𝑖𝑎𝑔(𝜎02 𝜎𝑑1
2⁄ , 𝜎02 𝜎𝑑2
2⁄ , … , 𝜎02 𝜎𝑑𝑚1
2⁄ ) (for direction measurements), 𝐏𝐬 =
𝑑𝑖𝑎𝑔(𝜎02 𝜎𝑠1
2⁄ , 𝜎02 𝜎𝑠2
2⁄ , … , 𝜎02 𝜎𝑠𝑚2
2⁄ ) (for distance measurements), 𝜎02 : The variance of unit weight
is chosen as 0.009 mgon2, 𝑚1 and 𝑚2 : The number of direction measurements and the number of
distance measurements, respectively, all assumed to be uncorrelated.
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
5.2 GPS Network
Deformation network has 3 points which were established at the Davutpasa Campus of Yildiz
Technical University, 4 points from TUSAGA-AKTIF (SARY, KABR, SLEE, ISTN) network and
2 points from IGS network (ISTA, TUBI) as shown in Fig. 3. Thirty two baselines were observed in
the network. Approximate baseline distances of the points used in the process stage with reference
to ISTA varying from 14.2 to 99.8 km.
The network were processed at Bernese v5.0 software with observing points OBC1, OBC2 and
OBC3 (Dach et al., 2007). 7 hours measurements were considered at processing stage. The optimal
values were used as stated in Bernese v5.0 guide. Precise GPS orbits and Earth rotation parameters
(EOP) were obtained from IGS data center (International GNSS Service). Approximate coordinates
were calculated for new points and receiver errors by one point positioning with ionosphere free L3
frequency.
Fig. 3. GPS Network (Erdogan 2011, Erdogan & Hekimoglu 2014)
Cycle slips and outliers were checked by generating triple differences. Then, outliers were removed
and cycle slips were corrected or initial phase ambiguity parameter were added to cycle slips which
could have not corrected. Then using QIF (Quasi Ionosphere Free) technique initial phase
ambiguity parameters were solved. Tropospheric zenith delay was disposed using Saastamoinen
model. To use for defining the datum ISTA, which is IGS point and computed by SOPAC, was
chosen at the datum 2005.0 and epoch 2010.835.
Obtained three-dimensional Cartesian coordinates were transformed to local coordinate
(Topocentric coordinate) system. Since the applied GPS measurement complied with the criteria in
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
the paper written by Eckl et al., 2001, the weight matrix was obtained by using the standard
deviations given in this article.
6. RESULTS OF EXPERIMENTS
6.1 Results for Horizontal Control Network
The components of displacement circle were obtained for points A, B and C which are assumed as
displaced and radius was calculated as r = 29.9 mm. Then, the EMDD magnitudes were also
calculated, using the 2nd
Empirical Technique: Step Approach test. To compare the magnitude with
theoretical approach the minimum detectable displacements were computed at 40 different
directions from Eqs. (19) - (20) with α = 0.05 and β = 0.20. The displacements magnitudes were
obtained for α = 0.05 and β = 0.30 as well (Fig. 4, Fig 5, Fig. 6).
Fig. 4. TMDD and EMDD magnitudes for point A
“+” refers to displacement magnitudes which were obtained through 2nd
Empirical Technique: Step
Approach Testing whose minimum detectable displacements magnitudes of each points were
calculated and shown for at 10 grade intervals.
“*” and “°” shows the TMDD magnitudes when α=0, β=0.20 and α=0, β=0.30, respectively.
The field between two circles at Fig. 4, Fig. 5, Fig. 6 are corresponds to (r, 2r) interval that is
obtained from 1st Empirical Technique: Using Displacement Ellipse shown as sect. 3.1. This
algorithm was used for calculating the radius of circles.
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
Fig. 5. TMDD and EMDD magnitudes for point B
Fig. 6. TMDD and EMDD magnitudes for point C
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
The calculated TMDD magnitudes are greater than the displacements obtained from the empirical
approaches. The TMDD magnitude is getting closer to EMDD magnitude when β=0.30.
6.2 Results for GPS Network
The up component was not considered in the GPS network processing. The results were calculated
for two dimensional network (n = north, e = east). The components of the displacement circle were
obtained for point OBC1, OBC2 and OBC3 which are called object points and the radius was
computed as r=7.1 mm. Then, the EMDD magnitudes were also calculated, using the 2nd
Empirical
Technique: Step Approach Testing.
To compare the magnitude with theoretical approach the minimum detectable displacements were
computed at 40 different directions from Eqs. (19) - (23) with α = 0.05 and β = 0.20. The
displacements magnitudes were obtained for α = 0.05 and β = 0.30 as well (Fig. 7, Fig 8, Fig. 9).
Fig. 7. TMDD and EMDD magnitudes for point OBC1
The field between two circles at Fig. 7, Fig. 8 and Fig. 9 are corresponds to radius (r, 2r) interval
that were obtained from 1st Empirical Technique: Using Displacement Ellipse shown as sect. 3.1.
This algorithm was used for calculating the radius of circles.
“+” refers to displacement magnitude that were obtained by the 2nd
Empirical Technique: Step
Approach Testing whose minimum detectable displacements magnitudes of each points were
calculated and shown for at 10 grade intervals.
mm (east)
mm
(n
ort
h)
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
“*” and “°” shows the TMDD magnitudes for α=0.05 and β=0.20, α=0.05 and β=0.30, respectively.
Fig. 8. TMDD and EMDD magnitudes for point OBC2
Fig. 9. TMDD and EMDD magnitudes for point OBC3
mm
(n
ort
h)
mm (east)
mm
(n
ort
h)
mm(east)
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
The calculated TMDD magnitudes are greater than the displacements obtained from the empirical
approaches. The TMDD magnitude is getting closer to EMDD magnitude when β=0.30 for GPS
network.
7. CONCLUSIONS
The MDD is very important for design of the deformation monitoring network. This value
classically are estimated based on the α = 0.05 and β = 0.20. However, since it is a theoretical value,
magnitude of TMDD changes depending on the α and β. In this paper, two different EMDD
techniques have been shown for estimating the MDD. The magnitudes of the TMDD techniques are
greater than the ones of the EMDD techniques. Accordingly, the simulation of the displacement
should be based on EMDD techniques for the reliability estimation of the deformation analysis
method. Otherwise, obtained result values from TMDD can be rather optimistic than most probable
values.
ACKNOWLEDGEMENTS
First Author thanks to the Scientific and Technological Research Council of Turkey for PhD. thesis
grant. GPS network based on the International GNSS Service (IGS) stations is used to simulate the
GNSS network. Also, GPS data were downloaded from the Scripps Orbit and Permanent Array
Center (SOPAC). Authors are very grateful to IGS for GNSS data and products. IGS using SOPAC
servers we are gratefully acknowledge to SOPAC as well.
REFERENCES
Aydin, C., & Demirel, H., 2005. Computation of Baarda’s lower bound of the non‐centrality
parameter. Journal of Geodesy, 78(7‐8), pp. 437‐441.
Aydin, C., Arslan N., and Demirel, H., 2004. Deformasyon Analizinde Duyarlık, HKMO Dergisi,
90, pp. 12-18.
Betti, B., Biagi, L., Crespi, M., and Riguzzi, F., 1999. GPS sensitivity analysis applied to permanent
deformation control networks. Journal of Geodesy, Berlin, 73(3), pp. 158-167.
Cai, J., Wang, J., Wu, Jicang., Hu, C., Grafarend, E., and Chen, J., 2008. Horizontal deformation
rate analysis based on multiepoch GPS measurements in Shanghai. J. Surv. Eng., 134(4), pp.
132–137.
Caspary, W., Chen, Y.Q. and Köning, R., (1983). Kongruenzuntersuchungen in
Deformationsnetzen durch Minimierung der Klaffungsbeträge. Schriftenreihe des
wissenscheftlichen Studienganges Vermessungswesen an der Hochschule der Bundeswhr, 9, pp.
77-94.
Caspary, W., 1988. A robust approach to estimating deformation parameters. Proc. of the 5th
Canadian Symp. on Mining Surveying and Rock Deformation Measurements, A. Chrzanowski
and W. Wells, eds., Dept. of Surveying Engineering Univ. of New Brunswick, Fredericton,
N.B., pp. 124–135.
Caspary, W.F., Haen, W., and Borutta, H., 1990. Deformation analysis by statistical methods.
Technometrics, 32(1), pp. 49-57.
Cooper, M.A.R., 1987. Control surveys in civil engineering. Collins, London.
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
Dach, R., Hugentobler, U., Fridez, P., and Meindl, M., 2007. Bernese GPS Software Version 5.0
Draft, Astronomical Institute, University of Berne, Switzerland.
Dogan, U., 2007. Accuracy Analysis of Relative Positions of Permanent GPS Stations in the
Marmara Region, Turkey. Survey Review, 39(304), pp. 156‐165.
Duchnowski, R., & Z. Wisniewski, 2014. Comparison of two unconventional methods of estimation
applied to determine network point displacement. Survey Review, 46, pp. 401-405.
Eckl, M.C., Snay, R.A., Soler, T., Cline, M.W., and Mader, G.L., 2001. Accuracy of GPS‐derived
relative positions as a function of inter‐station distance and observing‐session duration. Journal
of Geodesy, 75, pp. 633‐640.
Erdogan, B. 2011. Investigation of the reliability of deformation analysis for different methods,
PhD thesis, Department of Geomatic Engineering, Yildiz Technical Univ., Istanbul, Turkey(in
Turkish).
Erdogan B., & Hekimoglu S., 2014. Effect of subnetwork configuration design on deformation
analysis. Survey Review, 46(335), pp. 142-148.
Even‐Tzur, G., 2006. Datum Definition and its Influence on the Reliability of Geodetic Networks.
Zeitschrift für Vermessungwesen, 2, pp. 87‐95.
Hekimoğlu Ş., Erdoğan B., Butterworth S., 2010. Increasing the Efficacy of the Conventional
Deformation Analysis Methods: Alternative Strategy, J. Surv. Eng., 136(2), pp. 53‐62.
Koch, K.R., 1985. Ein statistisches auswerteverfahren für deformationsmessungen. Allgemeine
Vermessung‐Nachrichten. 92(3), pp. 97‐108.
Niemeier, W. 1985. Deformations analyse. in: H. Pelzer, (ed.), Geodaetische Netze in Landes – und
Ingenieurvermessung II, Konrad Wittwer: Stuttgart, pp. 559‐623
Nowel, K. & Kaminski W., 2014. Robust estimation of deformation from observation differences
for free control networks. Journal of Geodesy, 88, pp. 749–764.
Nowel K., 2015. Robust M-Estimation in Analysis of Control Network Deformations: Classical and
New Method. J. Surv. Eng., 141(4), pp. 1-10.
Pelzer, H., 1971. Zur analyse geodätischer deformationsmessungen. Deutsche Geodätische
Kommission, München, Germany, C‐164.
Schaffrin, B., 1986. New estimation/prediction techniques for the determination of crustal
deformations in the presence of prior geophysical information. Tectonophysics, 130(1–4), pp.
361–367.
Schaffrin, B., & Bock, Y. 1994. Geodetic deformation analysis based on robust inverse theory.
Manuscr. Geod., 19(1), pp. 31–44.
Snow, K., & Schaffrin, B., 2007. GPS-network analysis with BLIMPBE: An alternative to least-
squares adjustment for better bias control. J. Surv. Eng., 133(3), pp. 114–122.
Soler, T., Michalak, P., Weston, N.D., Snay, R.A., and Foote, R.H., 2005. “Accuracy of OPUS
solutions for 1‐ to 4‐h observing sessions. GPS solutions”,10, pp. 45‐55.
Sanli, D.U. & Engin, C., 2009. Accuracy of GPS Positioning Over Regional Scales. Survey Review,
41(312), pp. 192‐200.
Velsink, H., 2015. On the deformation analysis of point fields. Journal of Geodesy, Vol. 89(11), pp.
1071-1087.
Welsch, W., M. & Heunecke, O., 2001. Models and Terminology for the analysis of geodetic
monitoring observations, in: A. Chrzanowski, (ed.), Orange, California, USA, 390–412.
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017
BIOGRAPHICAL NOTES
Bahattin ERDOGAN is an Associate Professor in the Geomatic Engineering Department at the
Yildiz Technical University (YTU) in Istanbul, Turkey. He holds MSc and PhD in Geomatic
Engineering Department from the YTU. His research area mainly covers Robust Statistics,
Statistical Analysis, Geodetic Deformation Analysis and Satellite Positioning.
Serif HEKIMOGLU is a retired Professor in the Geomatic Engineering Department at the Yildiz
Technical University (YTU) in Istanbul, Turkey. He was retired in February, 2014. He holds MSc
in Geomatic Engineering Department from the YTU and PhD in Institute for Theoretical Geodesy
in Bonn. His research area mainly covers Robust Statistics, Statistical Analysis and Geodetic
Deformation Analysis.
Utkan Mustafa DURDAG is a PhD candidate in Geomatic Engineering and also works as a
research assistant at the Department of Geomatic Engineering in Yildiz Technical University
(YTU), Istanbul. He holds MSc and BSc degrees in Geomatic Engineering from the YTU. He is
now pursuing a Ph.D. in kinematic deformation analysis. His main research interests are; geodetic
deformation analysis, statistical analysis.
CONTACTS
Utkan Mustafa DURDAG
Department of Geomatic Engineering
Yildiz Technical University
34220 Istanbul, Turkey
Tel. +90 212 383 52 87
E-mail: [email protected]
Website: http://yarbis.yildiz.edu.tr/umdurdag/en
Theoretical and Empirical Minimum Detectable Displacements for Deformation Networks (8590)
Bahattin Erdogan, Serif Hekimoglu and Utkan Mustafa Durdag (Turkey)
FIG Working Week 2017
Surveying the world of tomorrow - From digitalisation to augmented reality
Helsinki, Finland, May 29–June 2, 2017